LCGC Europe - July 2001

LC troubleshooting

Buffers and Baselines

Nan S. Wilson, Ryan Morrison, LC Resources Inc., McMinnville, Oregon, USA and John W. Dolan, LC Resources Inc., Walnut Creek, California, USA.

Gradient baselines dont always behave as desired.

Drifting baselines are common with gradient elution liquid chromatography (LC) separations. Drift is influenced by the mobile phases organic solvents, the detection wavelength and the presence of additives. This months LC Troubleshooting focuses on the baseline characteristics of several common LC buffers and methods to minimize drift. When comparing figures in this column, readers should note the scale of the y axis, because it is not the same in all figures. For example, the absorbance scales for Figures 1 and 2 differ by 10-fold. Also, dont place too much emphasis on the peaks in the various runs we chose the chromatograms from our files to illustrate buffer and solvent drift and we made no special efforts to remove extraneous peaks. Extra peaks in a gradient run are often the result of contaminants in the buffer or water and they can be a challenge to eliminate. An earlier LC Troubleshooting column has an example of the process to isolate extraneous peaks (1). Phosphate An Old Friend Phosphate buffers have long been a favourite of chromatographers. Some of phosphates endearing qualities include its low cost, high purity, convenient preparation, useful pH ranges and good chromatographic behaviour. Figure 1 shows blank gradients for a 10 mM phosphate buffer with methanol as the organic solvent. At a low wavelength of 215 nm, the baseline drift is minimal and a flat baseline is produced at 254 nm. We observe drift when the UV absorbance of the starting solvent does not match that of the ending solvent; however, as long as the drift is small enough to allow peaks to stay

Absorbance (AU)

Absorbance (AU)

on scale and provide accurate integration, it can be ignored. Workers should be careful about the solubility of buffer salts in the organic solvent, especially when using phosphate buffers. For this reason, our laboratory personnel avoid using phosphate buffers at concentrations greater than approximately 10 mM combined with organic concentrations greater than approximately 80%. Methanol is less problematic in this regard than acetonitrile. It is best to maintain constant buffer concentration during the run with equimolar buffer in both the A and B solvents. However, many workers take the easy way out and run with buffer in the A bottle and organic solvent in the B bottle, as is the situation for Figure 1. Although this causes a gradient in buffer strength, it seldom is a problem if the buffer concentration is greater than 10 mM, organic concentrations are less than approximately 80%, and the buffer is used within its normal buffering range. For reproducible separations, it is important to specify in the LC method how the buffer is

prepared and the formulation of the A and B solvents. Figure 2 illustrates the effect of a B solvent with higher absorbance than the A solvent. In this instance, the same phosphate buffer as in Figure 1 was used with tetrahydrofuran instead of methanol as the organic solvent. As Figure 2 shows, tetrahydrofuran causes the baseline to drift to two absorbance units (2 AU) at 215 nm. This drift makes tetrahydrofuran unusable as an organic solvent for low-wavelength gradients. Although many modern LC detectors are linear to an upper limit of 2 AU, remember that the 2 AU linear range includes any background. For the present example, 50% tetrahydrofuran generates a 1 AU background, so only 1 AU is available as the maximum peak height to maintain detector linearity. Most older UV detectors and many older data systems are limited to a 1 AU maximum detector signal, which further limits the usefulness of tetrahydrofuran as a solvent.

2.0

0.10 0.00 0 2 4 6 8 Time (min) 10

1.0

0.0 0 2 4 6 8 Time (min) 10 12

Figure 1: Baselines obtained using phosphate methanol gradients. Solvent A: 10 mM potassium phosphate (pH 2.8); solvent B: methanol; gradient: 580% solvent B in 10 min. Detection wavelength: 215 nm (upper trace), 254 nm (lower trace).

Figure 2: Baselines obtained using phosphate tetrahydrofuran gradients. Solvent A: 10 mM potassium phosphate (pH 2.8); solvent B: tetrahydrofuran; gradient: 580% solvent B in 10 min. Detection wavelength: 215 nm (upper trace), 254 nm (lower trace).

LC troubleshooting

LCGC Europe - July 2001

At higher wavelengths, such as the 254 nm run shown in Figure 2, tetrahydrofuran is satisfactory, and it may be the solvent of choice because of its separation characteristics. Trifluoroacetic Acid Another Favourite An alternative buffer for low-pH work is trifluoroacetic acid. Trifluoroacetic acid is a favourite buffer for the separation of biomolecules. Tetrahydrofuranacetonitrile gradients, such as those shown in Figure 3, represent standard conditions for biomolecule separations. The volatility of trifluoroacetic acid makes it compatible with mass spectrometry (MS) detection or suitable for sample recovery after mobilephase evaporation. Because of the ease of formulation (just add 1 mL/L) and its ability to minimize absorbance differences between the A and B solvents at low wavelengths, 0.1% trifluoroacetic acid can be added to water as the A solvent and acetonitrile as the B solvent. Figure 3 also illustrates the effect of the system dwell volume and the column volume on the baseline. The combination of these two volumes delays the arrival of
Absorbance (AU)

the gradient at the detector by almost 3 min in this instance, as the baseline step for the 215 nm plot shows. Note that although the step looks large in Figure 3, the actual offset is approximately 0.03 AU. The column dead time can be identified by the baseline disturbance immediately after 1 min. Acetate for Intermediate pH Trifluoroacetic acid is useful near pH 2, and phosphate is useful from approximately pH 2 to pH 3.1 and from pH 6.2 to pH 8.2. For intermediate pH values, acetate buffers are often the best choice. These buffers also have the advantage of volatility, so they can be used with LCMS detection. However, using acetate can be a challenge if chromatographers fail to take proper care during mobile-phase formulation. This problem is illustrated in Figure 4, in which we took the buffer-in-A solvent and organic-in-B solvent approach, as was the situation with phosphate in Figures 1 and 2. At 215 nm, ammonium acetate has much greater background absorbance than methanol and results in the severely negative drift in Figure 4.

Absorbance (AU)

0.10

0.2 0.1 0.0 0.1 0.2 0 10 20 30 Time (min) 40

0.00 0 2 4 6 Time (min) 8 10

Negative drift is especially problematic because most data systems cannot process detector outputs less than roughly 100 mV (equivalent to 0.1 AU in this situation). Under these restrictions, analysts would observe the baseline dropping to 0.1 AU and then remaining flat, much as in the days of strip chart recorders when the pen would drift to the bottom of the paper. This set of conditions is clearly unacceptable. To compensate for the negative drift, workers can add a UVabsorbing compound to the B solvent so its absorbance matches the A solvent. In the present situation, we accomplished this task by adding ammonium acetate to the methanol. For the best performance with ammonium acetate, chromatographers should prepare the A and B solvents so they have the same buffer concentration 25 mM in this instance. When we used equimolar buffer in both solvents, the drift was greatly reduced, as Figure 5 shows (note the absorbance scale change). We observed a slight positive drift, which is easy for the data system to handle. The magnitude of the drift (approximately 0.2 AU in this situation) is small enough so integration will be satisfactory and the chromatograms will be visually pleasing. Note that when working at the higher wavelength of 254 nm, matching the mobile-phase absorbance is unnecessary. High-pH Buffers Although phosphate is a useful buffer at pH levels as high as pH 8.2, we recommend avoiding phosphate at levels higher than pH 8 in favour of organic buffers that minimize the dissolution of the silica-based column packing (2). One alternative is Tris (tris[hydroxymethyl]aminomethane), a buffer widely used by biochemists. Some workers have found ammonium bicarbonate to be especially useful at high pH levels (3).
1.0 0.0 0.1 0.2 0 2 4 6 Time (min) 8 10

Figure 3: Baselines obtained using trifluoroacetic acidacetonitrile gradients. Solvent A: 0.1% trifluoroacetic acid in water (pH 1.9); solvent B: 0.1% trifluoroacetic acid in acetonitrile; gradient: 1090% solvent B in 10 min. Detection wavelength: 215 nm (upper trace), 254 nm (lower trace).

Figure 5: Baselines obtained using equimolar ammonium acetatemethanol gradients. Solvent A: 25 mM ammonium acetate (pH 4) in 5% methanol; solvent B: 25 mM ammonium acetate in 80% methanol; gradient: 0100% in 40 min. Detection wavelength: 215 nm (upper trace), 254 nm (lower trace).

Absorbance (AU)

Absorbance (AU)

0.1 0.0 0.1 0.2 0.3 0 5 10 15 Time (min) 20

10

20 30 Time (min)

40

Figure 4: Baselines obtained using ammonium acetatemethanol gradients. Solvent A: 25 mM ammonium acetate (pH 4); solvent B: 80% methanol in water; gradient: 5100% solvent B in 40 min. Detection wavelength: 254 nm (upper trace), 215 nm (lower trace).

Figure 6: Baselines obtained using Trismethanol gradients. Solvent A: 50 mM Tris; solvent B: methanol; gradient: 580% solvent B in 20 min. Detection wavelength: 215 nm (upper trace), 254 nm (lower trace).

Figure 7: Baselines obtained using ammonium bicarbonatemethanol gradients. Solvent A: ammonium bicarbonate (pH 9); solvent B: methanol; gradient: 560% solvent B in 10 min. Detection wavelength: 215 nm (upper trace), 254 nm (lower trace).

Absorbance (AU)

0.0 0.2 0.4 0.6 0.8 1.0

LCGC Europe - July 2001

LC troubleshooting

Figures 6 and 7 illustrate examples of baselines obtained with these buffers. In both situations, the present negative drift is sufficient to cause problems with data-system performance at low wavelengths. At higher wavelengths, drift is no problem. Conclusions Examining gradient baselines for several combinations of buffers and organic solvents is useful to illustrate several common problems and practices in mobilephase preparation. As a general rule, baseline drift caused by absorbance mismatch between the solvents will be more problematic at lower wavelengths than at higher ones. To avoid undesirable drift, workers may need to add a UV-absorbing compound to one solvent or the other so their absorbance is similar. From a scientific standpoint, analysts will get the best chromatographic behaviour and reproducible separations if the A and B solvents have the same concentration of buffer and the organic solvent concentration varies during the gradient. From a practical standpoint, however, workers can often get satisfactory results by placing buffer in only the A solvent. Although we did not discuss it in this column, using buffers within their effective buffering range 1 pH unit from their pKa value is important. Outside this range, chromatographers will obtain poor buffering even with high buffer concentrations. The runs described in this column also illustrate the importance of running a blank gradient to determine drift problems and to identify any background peaks in the gradient. The baseline drift problems discussed in this column will not occur with isocratic separations using the same mobile-phase components. Any baseline offset caused by solvent or buffer absorbance will be hidden when the detector autozeros at the beginning of each run. However, the background absorbance of the mobile phase is still present, so the useful linear range of the detector may be reduced. References
(1) M.D. Nelson and J.W. Dolan, LCGC Int., 11(12), 764769 (1998). (2) L.R. Snyder, J.J. Kirkland and J.L. Glajch, Practical HPLC Method Development (John Wiley & Sons, New York, USA, 1997), 202. (3) Uwe Neue, personal communication, April 2001.

development of stability-indicating assays for pharmaceutical products. LC Troubleshooting editor John W. Dolan is president of LC Resources Inc., of Walnut Creek, California, USA, and a member of the Editorial Advisory Board of LCGC Europe. Direct correspondence about this column to LC Troubleshooting, LCGC Europe, Advanstar House, Park West, Sealand Road, Chester CH1 4RN, UK, e-mail: dhills@advanstar.com Readers can also direct questions to the on-line Chromatography Forum at http://www.chromatographyonline.com

Ryan Morrison and Nan Wilson are chemists at LC Resources Inc. (McMinnville, Oregon, USA) who have special interests in column characterization and the